On 28/11/2025 14:50, Alexander Gordeev wrote: > Would it make sense to explicitly describe the policy wrt sleeping while > in lazy MMU mode? If I understand the conclusion of conversation right: > > * An arch implementation may disable preemption, but then it is arch > responsibility not to call any arch-specific code that might sleep; > * As result, while in lazy MMU mode the generic code should never > call a code that might sleep;
I think that's a good summary, and I agree that the second point is not obvious from the comment in <linux/pgtable.h>. This series it not making any change in that respect, but I'll add a clarification in this patch (or a separate patch). - Kevin
